29 March 2016 Spatiotemporal-atlas-based dynamic speech imaging
Author Affiliations +
Abstract
Dynamic magnetic resonance imaging (DS-MRI) has been recognized as a promising method for visualizing articulatory motion of speech in scientific research and clinical applications. However, characterization of the gestural and acoustical properties of the vocal tract remains a challenging task for DS-MRI because it requires: 1) reconstructing high-quality spatiotemporal images by incorporating stronger prior knowledge; and 2) quantitatively interpreting the reconstructed images that contain great motion variability. This work presents a novel imaging method that simultaneously meets both requirements by integrating a spatiotemporal atlas into a Partial Separability (PS) model-based imaging framework. Through the use of an atlas-driven sparsity constraint, this method is capable of capturing high-quality articulatory dynamics at an imaging speed of 102 frames per second and a spatial resolution of 2.2 × 2.2 mm2. Moreover, the proposed method enables quantitative characterization of variability of speech motion, compared to the generic motion pattern across all subjects, through the spatial residual components.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Maojing Fu, Jonghye Woo, Zhi-Pei Liang, Bradley P. Sutton, "Spatiotemporal-atlas-based dynamic speech imaging", Proc. SPIE 9788, Medical Imaging 2016: Biomedical Applications in Molecular, Structural, and Functional Imaging, 978804 (29 March 2016); doi: 10.1117/12.2216528; https://doi.org/10.1117/12.2216528
PROCEEDINGS
9 PAGES


SHARE
Back to Top